Why Depensee has its own integration library

Depensee starts from a simple product idea: show a project as a map, where tasks become nodes and the relations between them become edges. But most teams already have a tracker: Jira, Yandex Tracker, Asana, or something else. So Depensee should not replace them on day one. It should read and normalize data from different systems carefully.

That is why integrations live in a separate library - trackerkit.

Why a separate layer

Tracker APIs look similar only from a distance. Jira has issues, projects, issue links, and a configurable hierarchy model. In Yandex Tracker, tasks do not live in “projects” in the usual planning-product sense: they live in queues, and a queue is the closest match for what Depensee calls a project. Asana has its own set of entities again: workspaces, projects, tasks, dependencies, and subtasks.

If all those differences leak directly into the Depensee backend, product logic quickly turns into a pile of if jira, if yandex, if asana. trackerkit keeps provider details behind one contract.

The less obvious parts

The clearest example is task relations.

In Jira, relations are highly customizable. A team may have its own link types, inward/outward labels, and separate mapping rules for blocks, relates, or a structural relation like contains. That is why trackerkit supports explicit Jira relation mapping.

Yandex Tracker is stricter and more direct: it has native links between issues, including dependencies and parent-child relations. But it has a different surprise - the API-level “project” is not the project you usually expect in planning software. For task work, the main container is a queue, so trackerkit normalizes queues as Project in the shared model.

Why it matters

Depensee is built around the map, not around one specific tracker. trackerkit keeps that boundary clean: the backend works with tasks and relations in one shape, while the library translates from each provider’s language.

That makes it easier to add new trackers, test integrations, and keep the Depensee product model separate from external API details.

trackerkit is still an early library, but it is the layer that will support the MVP import and read-side sync from Jira and Yandex Tracker. The code is open on GitHub: github.com/depensee-dev/trackerkit.